The mediators of real-world executive functions in autism during the transition into adulthood are mainly unknown. This study aimed to identify the mediators for the behavioral and cognitive domains of real-world executive functions in late adolescent and young adult autistic populations. We followed up 289 autistic children (aged 11.6 ± 3.8, male 82.2%) and 203 non-autistic controls (10.7 ± 2.9, 66.5%) to their ages of 16.9 ± 4.7 and 15.8 ± 3.9, respectively. The Behavior Rating Inventory of Executive Function, comprising two global scales (Behavioral Regulation Index and the Metacognitive Index), was used to assess real-world executive functions at Time 2. Mediators of behavior regulation and metacognition were explored among a wide range of clinical correlates and parental bonding. We found that the autistic group had lower executive functions than the non-autistic group with lower behavior regulation and metacognition subscores. Several childhood factors were significant mediators. Multiple mediation analyses revealed that childhood inattention remained a significant mediator for both behavior regulation and metacognition in late adolescence, and peer problems were the specific mediators of metacognition. Our findings suggest several childhood factors may mediate real-world executive functions during late adolescence. Treating common mediators such as inattention may improve real-world executive functions in autistic individuals during adulthood. Lay abstract Childhood factors that predict real-world executive function in autism spectrum disorder during the transition into adulthood are largely unknown. This study aimed to identify the predictors for the behavioral and cognitive aspects of real-world executive function in late adolescent and young adult autistic populations. We followed up 289 autistic youth (mean age 11.6 years) and 203 non-autistic controls (10.7 years) to their ages of 16.9 and 15.8, respectively. The Behavior Rating Inventory of Executive Function scale was used to measure the real-world executive function at late adolescence and young adulthood at follow-up. Potential predictors such as autistic symptoms, inattention or hyperactivity symptoms, peer relationship, emotional symptoms, and parenting styles were assessed in childhood at first enrollment. The results showed that childhood inattention, withdrawn behaviors, social communication difficulties, and child-reported emotion and inattention/hyperactivity may predict real-world lower executive function in late adolescence and young adults with autism. When separating executive function into behavioral and cognitive aspects, we found that oppositional behaviors and peer problems were specific predictors for behavioral regulation and cognitive function, respectively. Our findings suggested that treating common predictors in childhood, such as inattention, may potentially improve real-world executive function in autism during the transition into adulthood.